Vett

Provides a registry and execution environment that scans AI agent skills for malicious code, credential theft, and data exfiltration before they run and isolates agents to prevent cross-agent access. Sells to developers, IT and operations teams in businesses that deploy AI coding agents such as Claude Code, Cursor, and GitHub Copilot. Delivered as a SaaS registry and CLI that signs verified skills with Sigstore and as a hosted Enclave that runs each agent in its own Firecracker microVM with vaulted credentials and an audit trail.

Key features

• Skill scanning with 40+ static analysis rules
• AST-level injection and exfiltration detection
• Sigstore signing for verified skills
• Firecracker microVM per-agent isolation
• Dedicated kernel and memory per agent
• Vaulted credentials with scoped tokens
• Immutable audit trail with replay
• MCP gateway for tool traffic
• Dependency CVE and typosquatting checks
• Support for 46+ AI agents